Federated learning (FL) serves as a distributed machine learning framework that addresses the challenges of data silos while preserving data privacy. Specifically, FL enables multiple participants to collaboratively train a global model by sharing local updates without exposing their raw local data. Although FL achieves physical data isolation through local update sharing mechanisms, it still faces emerging security threats. On the one hand, adversaries may reconstruct sensitive data features or infer client attributes by analyzing local updates. On the other hand, clients might upload malicious updates to disrupt global model aggregation, causing performance degradation. To solve these issues, we propose an anonymous and Byzantine-robust FL scheme with secure and efficient aggregation. First, we propose a single-masking protocol that not only preserves data privacy but also enhances aggregation efficiency. Second, we eliminate client message metadata, such as source IP addresses and timestamps, through secure shuffling, achieving client anonymity in conjunction with the single-masking protocol. Additionally, we implement a baffle mechanism to resist the impact of malicious updates on the global model, thereby ensuring Byzantine robustness. Security analysis demonstrates that our scheme simultaneously preserves data privacy and identity anonymity. Experimental results show that our scheme can effectively resist poisoning attacks, even if 50% of the fog nodes are contaminated by malicious clients. Moreover, the aggregation efficiency of the proposed scheme is improved by over 20%.

Laconic Private Set Intersection (LPSI) is a type of PSI protocols characterized by the requirement of only two-round interactions and by having a reused message in the first round that is independent of the set size. Recently, Aranha et al. (CCS'2022) proposed a LPSI protocol that utilizes the pairing-based accumulator. However, this protocol heavily relies on time-consuming bilinear pairing operations, which can potentially cause a bottleneck. Furthermore, in certain scenarios like contact tracing, it is sufficient to only reveal the intersection cardinality. To tackle this problem and expand on its functionalities, we introduce a cloud-assisted two-party LPSI cardinality (TLPSI-CA) that inherits the properties of LPSI. Interestingly, the cloud-assisted TLPSI-CA eliminates the direct interaction between the sender and receiver, enabling the sender's message to be reused across any number of protocol executions. Besides, we further extend it to the multi-party scenario, which also possesses laconic properties. Then, we prove the two protocols' security in achieving the defined ideal functionalities. Finally, we evaluate the performance of both protocols and find that TLPSI-CA successfully reduces the local computation costs for participants. Additionally, the multi-party protocol performs similarly to TLPSI-CA, with the exception of the higher communication costs incurred by the receiver.

Abstract With the rapid development of smart health, the privacy problem of medical data has become more prominent. Aiming at the problem of mining the potential value of medical data and realizing secure sharing, a fine‐grained medical data sharing scheme based on federated learning is proposed. The scheme uses collaboration‐oriented attribute‐based encryption technologies to formulate fine‐grained access strategies, allowing medical institutions or doctors to decrypt individually or collaboratively with certain conditions to achieve the purpose of accurately screening the required medical data. In the proposed scheme, the model parameters are shared such that the screened medical data is modeled and analyzed based on federated learning, which allows more people to enjoy top medical resources. In addition, a blockchain‐based incentive mechanism is used to reward medical institutions which are either honest with high‐quality or helpful in decryption. Hence, the enthusiasm of various medical institutions to screen data and participate in federal learning is improved. Finally, security analysis shows that the scheme is secure, and theoretical analysis and simulation test show the practicability of the scheme.

The rapid development of smart healthcare system in the Internet of Things (IoT) has made the early detection of many chronic diseases more convenient, quick, and economical. However, when healthcare organizations collect users’ health data through deployed IoT devices, there are issues of compromising users’ privacy. In view of this situation, this paper introduces federated learning technology to solve the problem of data security. In this paper, we consider the two main problems of federated learning applications in IoT smart healthcare system: (1) how to reduce the time overhead of system running and (2) how to authenticate that the user device uploading data is deployed by the system itself. To solve the above problems, we propose the first federated learning scheme based on full dynamic secret sharing. First, we use a two‐mask protocol to keep the user’s local model parameters confidential during federated learning. Then, based on homogeneous linear recursive equation, homomorphic hash function, and elliptic curve cryptosystem, the full dynamic secret sharing and user identity authentication are realized. In addition, our scheme allows users to join or quit during training. Finally, we have carried out simulation test on this scheme. The experimental results show that the efficiency of our scheme is improved by about 60% on average in the case of no user dropping and by about 30% in the case of some users dropping.

Blockchain systems face two emergent problems, namely scalability and privacy, each of which has been addressed independently. However, how to achieve privacy and scalability simultaneously remains a challenging problem for blockchains. In this article, we propose a privacy-preserving and scalable permissioned blockchain system called HyperMaze employing the zero knowledge proof technique and a hierarchical system architecture. It gains scalability by adopting a hierarchy of multiple blockchains that processes transactions in parallel. We design anID-based dual-balance account modelwhere an identity-based account is associated with two types of balances–a plaintext balance and a private (zero-knowledge) balance. Furthermore, we design a two-phase cross-chain transaction mechanism (2PXT) to achieve transaction privacy for both intra-chain and cross-chain transactions. We rigorously formulate a security model for HyperMaze under the universal composability framework, and then provide a simulation-based security proof. A prototype of HyperMaze is implemented and a series of experiments are conducted over up to 2,600 nodes to evaluate its performance. The experimental results show that a 4-level, (7,10)-threshold, 6-ary HyperMaze system can reach 19,440 TPS and the transaction confirmation latency is only 9.5 seconds. To our best knowledge, HyperMaze is the first high-throughput privacy-preserving blockchain whose throughput is over 19,000 TPS.

Internet-of-Things enables interconnection of billions of devices, which perform autonomous operations and collect various types of data. These things, along with their generated huge amount of data, need to be handled efficiently and securely. Centralized solutions are not desired due to security concerns and scalability issue. In this article, we propose HIBEChain, a hierarchical blockchain system that realizes scalable and accountable management of IoT devices and data. HIBEChain consists of multiple permissioned blockchains that form a hierarchical tree structure. To support the hierarchical structure of HIBEChain, we design a decentralized hierarchical identity-based signature (DHIBS) scheme, which enables IoT devices to use their identities as public keys. Consequently, HIBEChain achieves high scalability through parallel processing as blockchain sharding schemes, and it also implements accountability by use of identity-based keys. Identity-based keys not only make HIBEChain more user-friendly, they also allow private key recovery by validators when necessary. We provide detailed analysis of its security and performance, and implement HIBEChain based on Ethereum source code. Experiment results show that a 6-ary, (7,10)-threshold, 4-level HIBEChain can achieve 32,000 TPS, and it needs only 9 seconds to confirm a transaction.